In this paper, visual servoing technique for a robotic manipulator to grasp a stationary or moving object is investigated. For this, a hand-held camera is used to measure the position of the robotic end-effector relative to an object, to derive feedback signals for the control of a robotic manipulator.
A visual servoing algorithm proposed is as follows:
First, image frame is scanned to find where the object is. After finding object, features are extracted from the picture, and the robot is moved according to this information. Secondly, the actual position of the object is compared with the desired reference position. If the object is not in that position, the robot is further moved. In case, the actual and the desired reference position of the object are equal, action of grasping the object is executed.
The proposed visual servoing algorithm is implemented on the PUMA 560 industrial robot and is experimentally demonstrated.